Introduction to Data Visualisation¶

ONS / NISR
2021

What is data visualisation¶

Napolean's march on Moscow.
Charles Joseph Minard (1869)

Why do data visualisation?¶

Amplifies Your Messaging¶

Pictures and visualizations of any kind have a more powerful impact on their audience than written words. This is because mentally, visuals are more strongly tied to memory. Visuals also reiterate your message in a unique way, strengthening their impact on your audience.

Data visualization empowers you to more effectively achieve your organization’s goals by tapping into these pictorial strengths. Presenting the results of your data analysis process in this more powerful and persuasive medium allows you to amplify your messaging to internal (and external) stakeholders, more easily uniting everyone behind a common organizational story.

Why It’s Critical: Amplifying your organizational story isn’t accomplished with just any form of data visualization, though. In order to be most effective, you need to utilize the correct chart or display for the situation. Choosing the wrong data visualization might overwhelm or confuse your audience – achieving the opposite of your intended result.

Data Visualization Provides Clearer Understanding¶

Not only does data visualization help communicate your organization’s story to internal and external audiences, but it also helps you understand your own story better. Better, in fact, than any other data analysis tool.

By allowing you to process a large amount of information at once, data visualization opens windows of understanding into the workings and operations of your business or government agency, either by measuring impact or providing visual insight.

Why It’s Critical: Key decision-makers in your organization aren’t always experts in every activity of the enterprise, but with the right data visualization comparisons, these leaders are able to make quicker, more informed decisions. Just as before, beware of improper comparative visualizations – if your organizational data analysis is unclear, confusing or difficult to compare, your visualizations might be doing more harm than good.

Data Visualization Aids Decision Analysis¶

You don’t make decisions in a vacuum. Instead, your decision analysis is fueled by the data and information you have available. When you feed accurate, unbiased data visualizations into your decision-making tools, you have the ability to make better decisions for your enterprise.

However, the essential factor is that the data should be unbiased to allow for informed decision-making. Proper data visualizations don’t distort the underlying information with deceptive displays. In addition, charts and displays should be dynamically and consistently updated with the latest information so that their decision-making utility is kept relevant.

Why It’s Critical: Biased or “spun” information in your data visualizations could be costing your organization significantly. Often, employees and low-level managers are afraid of delivering bad news to their superiors, fearing that the news might unfairly suggest poor performance – or might incur punishment. As a result, these team members present deceptive or biased data visualizations that consistently show good news, even when the deep-down facts say otherwise.

In a 2012 paper entitled “When Is Good News Really Good News?”, Thomas Schleicher found that mid-level managers often cherry-pick favorable data points or resort to vague language when reporting bad news to their superiors. Schleicher discovered that this selectivity and avoidance had significant consequences:

Misreporting data or skewing data visualizations resulted in poor decisions by superiors who had incorrect or missing information. Organizations that didn’t report data accurately had overall poorer financial performance than those organizations with correct information. Ultimately, masking hard truths had dangerous effects on every enterprise in the study. Data visualization is tremendously powerful, and how you utilize it in your organization or business has drastic consequences – in either a positive or negative way. Improper data visualization taints your decision-making and muddles your message, but correctly applied data visualization makes everything you do more effective: powerful messaging, clearer insights and decisions you can stand behind.

Exploratory vs Explanatory visualisation¶

Exploratory: Exploratory data visualizations are appropriate when you have a whole bunch of data and you’re not sure what’s in it.

Explanatory: Explanatory data visualization is appropriate when you already know what the data has to say, and you are trying to tell that story to somebody else. It could be the head of your department, a grant committee, or the general public.

Exploratory Example¶

This is exploratory because we've not given the end user a specific story, instead we're inviting them to interact and interrogate the data themselves.

Exploratory Tools¶

Exploratory tools usually require some compute power behind them, so people can interact with the data. These can become costly to maintain and slow to run if the volume of users increases.

Explanatory Example¶

This is explanatory as we've highlighted a specific facet of our data that we want the audience to engage with.

Explanatory Tools¶

Lots of explanatory tools don't require any compute to sit behind them. They can usually be exported as images, or hosted in a web-page to retain some interactivity without having to dedicate resource.

What does good data visualisation look like?¶

Clarity¶

One of the key benefits of data visualisation is that it lets you explain your data in a visual language. But just because it's visual doesn't mean its is clearer.

Key tips for clarity¶

Simple beats complex every time.
The less information you can display and still tell the same story the better.

Consistency¶

Consistency in look and feel immediately creates an air of professionalism. With that comes trust which is incredibly important.

Charts from The Economist magazine
Various

Key tips for consistency¶

Use a house style to keep the same fonts and colours across all your charts
Look at other similar charts your organisation has made before, does yours differ? If so why? Does the difference make the chart better?
Show your work to other people in your organisation before you publish it. Ask for feedback, be prepared to make changes.

How do I choose the right visualisation?¶

Choosing the right visualisation easiest to do when you decide what aspects of your data you want to highlight.

Deviation¶

If you want to emphasise a variation (+ / -) from a fixed point or to show sentiment (positive / neutral / negative)

Correlation¶

If you want to show the relationship between two or more variables. Be aware, unless you state otherwise people will assume these correlations are causal.

Ranking¶

Use a ranking chart when an items position in an ordered list is more important than its absolute or relative value.

Distribution¶

Use to show how often values in a dataset occur. Showing the distribution or skew of a dataset can be very effective at conveying inequality.

Change over time¶

Use to emphasise trends, remember using the correct time scale is key for these charts.

Magnitude¶

Use to difference in sizes, either relative or absolute. Usually a countable number such as number of people, rather than a rate.

Part-to-whole¶

Use to show how a single value to can be broken down into many elements. Magnitude charts will be clearer if the user is only interested in total size.

Spatial¶

Use when the most important feature is the geographical pattern in the data.

Flow¶

Use to show the volume or intensity of movement between two or more states.

7 Tips for effective visualisation¶

Steal like an artist¶

Embrace the inspiriation you get from other people's work. If something look great, and would enhance the data you're working with steal the idea!

Avoid Clutter (Clarity)¶

Try and remove as much as you possibly can from your visualisation whilst keeping the message you're attempting to convey clear. *Clear and focused is always betters.

Use a grid layout¶

People will attempt read your dashboard just like any other document, usually left to right and top to bottom. Using a grid layout can make it much easier for someone to engage with your visualisation.

Think about your fonts¶

Fonts are a great way to highlight important information, guide the reader and create a professional look all at once. You can create a hierarchy with different font sizes. Try using colour to draw the readers eye to the most important information.

Keep your colour use simple¶

Simple colour schemes are usually more effective than having five or six different colours on the screen. This will simplify the visuals for your reader, allowing them to pick out the important information.

BANs (Big Ass Numbers)¶

Dont bury the key information. If there an important number or KPI, make it big and visible so anyone looking for it wont be able to miss it.

Collaborate and Iterate¶

No dashboard is going to be perfect on the first attempt. Iterate, remake the dashboard a few times until you've got something you're happy with, then test it!. Show it to other people, do they understand the message you're trying to convey? Repeat this process until you've got something great!

Data visualisation checklist¶

☑ The visualisation is clear and concise.

☑ The visualisation tells a story.

☑ The visualisation's data is appropriate for the intended audience.

☑ The visualisation is in the house style.

☑ The visualisation type is appropriate to for the data being shown.